The potential value of autostereoscopic display systems is well appreciated for a broad range of data visualization uses and for a wide range of applications that include entertainment, engineering, medical, government, security, and simulation fields. Autostereoscopic display systems include “immersion” systems, intended to provide a realistic viewing experience for an observer by visually surrounding the observer with a three-dimensional (3-D) image having a very wide field of view. As differentiated from the larger group of stereoscopic displays that include it, the autostereoscopic display is characterized by the absence of any requirement for a wearable item of any type, such as goggles, headgear, or special polarized or filter glasses, for example. That is, an autostereoscopic display attempts to provide “natural” viewing conditions for an observer.
An article entitled “3-D displays: A review of current technologies” by Siegmund Pastoor and Mathias Wopking in Displays 17 (1997) surveys various approaches that have been applied for obtaining autostereoscopic display images for one or more viewers. Among the many techniques described in the Pastoor et al. article are electro-holography, volumetric display, direction-multiplexed, diffraction-based, refraction-based, and reflection-based methods for autostereoscopic presentation. While each of these approaches may have merit in one or more specific applications, these approaches have a number of characteristic shortcomings that constrain usability and overall performance. As a group, these conventional approaches have been adapted for autostereoscopic displays, but allow only a narrow field of view and provide limited brightness and poor contrast. Imaging systems employing time-based or spatial multiplexing require complex image processing algorithms in order to provide left- and right-eye images in the proper sequence or with the necessary spatial separation. Time-based multiplexing introduces the inherent problem of image flicker. Spatial multiplexing generally produces an image having reduced resolution. Combining these multiplexing techniques, as is disclosed in European Patent Application 764 869 A2 (Ezra et al.), may provide an increased number of views, but does not compensate for these inherent drawbacks. A number of multiplexing technologies also require tracking of view eye position and compensation for changes in head position. As a further disadvantage, each of the imaging technologies described in the Pastoor et al. article present the viewer with a real image, rather than with a virtual image, a distinction described in more detail subsequently.
In an article entitled “An Autostereoscopic Display Providing Comfortable Viewing Conditions and a High Degree of Telepresence” by Klaus Hopf in IEEE Transactions on Circuits and Systems for Video Technology, Vol. 10, No. 3, April, 2000, a teleconferencing system employing a spherical mirror is disclosed, recommended particularly for its value in reducing chromatic aberration. However, the optical system disclosed in this article is subject to field curvature constraints, limiting its field of view. Notably, the system described in the Hopf article provides a virtual image; however, due to substantial field curvature, the total field of view of such a system is limited to less than about 15 degrees. While such a narrow field of view may be acceptable for videoconferencing applications, this level of performance would not be useful for a desktop display system.
Three-dimensional (3-D) imaging systems using curved reflective surfaces to form real images are known. For example, U.S. Pat. No. 3,447,854 (Minter) discloses a 3-D viewer that employs a curved mirror acting as a field lens for shifting the position of the viewing pupil for separate left-eye and right-eye images. Similarly, in an article entitled “Stereoscopic Display Using a 1.2-M Diameter Stretchable Membrane Mirror” by McKay et al., a large curved mirror is used in conjunction with left and right beamsplitters for providing a real image, shifting a convergence point for left- and right-image disparity to some position along the primary optical axis relative to the curved mirror surface. Significantly, with both the U.S. Pat. No. 3,447,854 apparatus and the McKay et al. apparatus, the projected image is focused onto the surface of the curved mirror itself. With this arrangement, since the projected image is focused onto the curved mirror; the mirror itself does not form the image, but simply directs light into the pupils of the viewer. Because the mirror serves as the display surface for this type of real image projection system, optimal viewing conditions and large field of view are obtained when using a large curved mirror placed a good distance away from the viewer.
These prior art imaging systems display a real image as contrasted with a virtual image. It is important to clarify the distinction between real and virtual imaging. A real image is defined as an image that is either focused on a surface or focused in an accessible location. That is, a screen can be placed at the position of a real image in order to display the image.
Unlike real imaging projection, a virtual imaging system forms an image that is not focused at an accessible location. That is, a virtual image is not formed by projection onto a display surface; if a display surface were positioned at the perceived location of a virtual image, no image would be formed on that surface. Instead, a virtual image is formed by an optical system. A virtual image can be considered to be formed by the eye itself, forming an image according to light incident on the retina. A virtual image occurs, for example, when an object is between the focal point and the vertex of a concave mirrored surface.
Virtual imaging provides an advantageous alternative to real image projection in some types of applications. U.S. Pat. No. 5,625,372 (Hildebrand et al.) outlines a number of inherent advantages of virtual imaging over the alternative real imaging that is commonly used for image projection. As one significant advantage for stereoscopic viewing, the size of a virtual image is not limited by the size or location of a display surface. Additionally, the source object for a virtual image may be small. A magnifying glass, as a simple example, provides an enlarged virtual image of a small object. Print viewed through a magnifying glass not only appears larger, it also appears to be located substantially behind the surface of the page where the print actually exists. By definition, then, a virtual image can exist at a location where no display surface exists. Thus, it can be seen that, in comparison with prior art systems that project a real image, a more realistic viewing experience can be provided by forming a virtual image that is formed so as to appear some distance away. Providing a virtual image also obviates the need to compensate for screen artifacts, as may be necessary when projecting a real image.
It is instructive to point out that the term “virtual image” is itself often casually misused in patent and related literature about immersive imaging systems and other apparatus that are often described as “virtual reality” systems. As one example, U.S. Pat. No. 5,976,017 (Omori et al.) makes a number of references to “virtual image” using the casual interpretation of this term as simply some type of electronically generated image, rather than using the definition understood by the optics practitioner. (Other examples of this unfortunate, casual use of the term “virtual image” can be easily found in the patent literature; as just a few additional examples, see U.S. Pat. No. 6,636,234 (Endo et al.) and U.S. Pat. No. 6,674,881 (Bacus et al.), both of which use the term “virtual image” in its casual sense, rather than holding to the definition used in optics.) In the disclosure of the present application, the term “virtual image” is used in its true optical sense.
It is generally recognized that, in order to minimize vergence/accommodation effects, a 3-D viewing system should display its pair of stereoscopic images, whether real or virtual, at a relatively large distance from the observer. For real image projection, this means that a large display surface must be employed, preferably placed a good distance from the observer. For virtual images, however, a relatively small curved mirror can be used as is disclosed in U.S. Pat. No. 5,908,300 (Walker et al.). The curved mirror in a virtual imaging system acts as a collimator, forming a virtual image that appears to be at a relatively large distance from the observer. In terms of focus, the image formed in such a virtual imaging system appears to be at infinity.
From an optical perspective, it can be seen that there would be advantages to autostereoscopic design using pupil imaging. A system designed for pupil imaging must meet a fairly demanding set of requirements, including the following:                (a) form separate images at left and right pupils correspondingly;        (b) provide the most natural viewing conditions possible, eliminating any need for goggles or special headgear;        (c) present the largest possible pupils to the observer, while limiting crosstalk between left and right views;        (d) allow the observer a reasonable freedom of movement;        (e) provide an ultra-wide field of view; and        (f) provide sufficient resolution for realistic imaging, with high brightness and contrast.        
It is recognized in the optical arts that each of these requirements, by itself, can be difficult to achieve. An ideal autostereoscopic imaging system must meet the challenge of each of these requirements to provide a more fully satisfactory and realistic viewing experience. Moreover, additional physical constraints presented by the need for a system to have small footprint, and dimensional constraints for interocular separation must be considered, so that separate images directed to each eye can be advantageously spaced and correctly separated for viewing. It is instructive to note that interocular distance constraints limit the ability to achieve larger pupil diameter at a given ultrawide field by simply scaling the projection lens.
Clearly, the value and realistic quality of the viewing experience provided by an autostereoscopic display system using pupil imaging is enhanced by presenting the stereo 3-D image with a wide field of view and large exit pupil. For fully satisfactory 3-D viewing, such a system should provide separate, high-resolution images to right and left eyes. To create a realistic illusion of depth and width of field, the observer should be presented with a virtual image that requires the viewer to focus at some distance.
It is well known that conflict between depth cues associated with vergence and accommodation can adversely impact the viewing experience. Vergence refers to the degree at which the observer's eyes must be crossed in order to fuse the separate images of an object within the field of view. Vergence decreases, then vanishes as viewed objects become more distant. Accommodation refers to the requirement that the eye lens of the observer change shape to maintain retinal focus for the object of interest. It is known that there can be a temporary degradation of a viewer's depth perception when the viewer is exposed for a period of time to mismatched depth cues for vergence and accommodation. It is also known that this negative effect on depth perception can be mitigated when the accommodation cues correspond to distant image position.
There are also other basic optical limitations for immersion systems that must be addressed with any type of optical projection that provides a wide field of view. An important limitation is imposed by the Lagrange invariant. A product of the size of the emissive device and the numerical aperture, the Lagrange invariant determines output brightness and is an important consideration for matching the output of one optical system with the input of another. Any imaging system conforms to the Lagrange invariant, whereby the product of pupil size and semi-field angle is proportional to the product of the image size and the numerical aperture. An invariant that applies throughout the optical system, the Lagrange invariant can be a limitation when using, as an image generator, a relatively small spatial light modulator or similar pixel array which operate over a relatively small numerical aperture, since the Lagrange value associated with the device is small. In practical terms, the larger the size of the image source, the easier it is to form an image having a wide field of view and large pupil.
In response to the need for more realistic autostereoscopic display solutions offering a wide field of view, commonly-assigned U.S. Pat. No. 6,416,181 (Kessler et al.), incorporated herein by reference and referred to as the '181 patent, discloses an autostereoscopic imaging system using pupil imaging to display collimated left and right virtual images to a viewer. In the '181 disclosure, a curved mirror is employed in combination with an imaging source, a curved diffusive surface, a ball lens assembly, and a beamsplitter, for providing the virtual image for left and right viewing pupils. Overall, the monocentric optical apparatus of the '181 disclosure provides autostereoscopic imaging with large viewing pupils, a very wide field of view, and minimal aberration.
While the autostereoscopic system of the '181 disclosure provides a high-performance immersive display, there is still room for improved embodiments. For example, while the '181 system provides a large viewing pupil, there would be advantages in obtaining even further increases in pupil size.
Generating its source image on a small spatial light modulator device, the '181 system overcomes inherent Lagrange invariant conditions by forming an intermediate curved image for projection on a curved diffusive surface. Use of the diffuser with the '181 apparatus is necessary because the image-forming device, typically a reflective liquid crystal on silicon (LCOS) or other spatial light modulator, is a relatively small image source device, measuring typically no more than about 1 inch square. At the same time, however, the use of a diffusive surface effectively reduces overall brightness, introduces some level of graininess to the image, and limits the achievable contrast.
Subsequent commonly-assigned applications have addressed the need for more compact autostereoscopic apparatus providing pupil imaging with virtual images and for achieving improved brightness levels. For example, U.S. patent application Ser. No. 10/662,208, cited above, discloses improved apparatus and methods for forming curved left and right intermediate images, using an approach that eliminates the need for use of a diffusive surface and allows the use of larger image sources which are able to provide additional brightness.
Although there is considerable interest in the enhanced display quality that true autostereoscopic virtual imaging affords, there remain a number of hurdles to widespread commercialization of compact, autostereoscopic displays suitable for desktop applications. There would be particular value in improvements that reduce cost and size and increase brightness, provide more natural viewing conditions, and enhance overall image quality.
Thus, it can be seen that there is a need for an improved autostereoscopic imaging apparatus that provides improved brightness, enhanced viewing pupil dimensions, reduced image aberrations, and higher resolution, at reduced size and cost.